Integrative Benchmarks | Models at Scale

The Brain-Score platform aims to yield accurate, machine-executable computational models of how the brain gives rise to the mind. We enable researchers to quickly get a sense of the alignment of their model(s) to currently dozens of neural and behavioral measurements, and provide these models to experimentalists to prototype new experiments and make sense of biological data.

Democratizing the Search for Scientific Models of Natural Intelligence

We believe in synergy between computational modeling and experimental brain and cognitive science. Brain-Score improves the efficiency of communication between experimentalists and modelers by providing experimental data as accessible benchmarks, and providing computational models in a unified form to experimentalists. All code is open-source and many community members have chosen to make their data or model weights fully public.

Collaborative

Join the active and growing Brain-Score community.

Open Source

Download and experiment. Completely free.

Accessible

Interact seamlessly with our curated leaderboard.

Leaderboard

At the core of Brain-Score is a leaderboard showing the latest evaluations of all available models on all available benchmarks. Models are ranked according to their average score across all the benchmarks because we believe the best model of a domain should explain all neural and behavioral data. The leaderboard also provides finer-grain per-benchmark and per-model scores for a more detailed view into where current models most need to improve on and which benchmarks are the most challenging.

Ready to Submit?

Below you will find links on how to prepare a model for submission and receive a Brain-Score! You will also find the same for submitting a benchmark, with detailed tutorials available for both modelers and experimentalists.